| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
Saint Louis University 1 School of Public Health
2 School of Medicine
3 College of Public Service, St. Louis, Missouri.
4 Saint Louis Veterans Administration Medical Center, St. Louis, Missouri.
5 Indiana University School of Medicine, Indianapolis.
6 Regenstrief Institute for Health Care, Indianapolis, Indiana.
7 Indianapolis Veterans Administration Medical Center, Indiana.
Address correspondence to Fredric D. Wolinsky, College of Public Health, the University of Iowa, 200 Hawkins Drive, E205 General Hospital, Iowa City, IA. E-mail: fredric-wolinsky{at}uiowa.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. Sense of control was measured at baseline and at each of six bimonthly follow-up interviews among 1,662 patients at two medical centers. Potential confounders were measured at baseline. Analyses include descriptive assessments of level and normative stability, repeated measures analysis of covariance, and hierarchical multiple linear and change score regressions.
Results. Although the sense of control is relatively stable between any two successive waves of data collection, significant gradual changes are observed over a 1-year period. Compelling evidence is found for statistically and substantively significant associations between age and the sense of control at baseline, and between age and changes in the sense of control over time. The only other major predictor of the sense of control is mental well-being.
Discussion. Longitudinal studies with repeated assessments over prolonged observation periods are now needed to elucidate age-related trajectories in the sense of control.
THE association between age and the sense of control is controversial (Fung, Abeles, & Carstensen, 1999
; Mirowsky, 1995; Rodin, 1986a, 1986b, 1987; Rodin, Timko, & Harris, 1985; Ross & Mirowsky, 2002), perhaps because of the inconsistency in its definition and measurement (Rodin, 1990; Skinner, 1996). The association, however, is critical. Individuals with a lower sense of self-control take less responsibility for their health, are less likely to engage in health protective behaviors, have inhibited immunologic function, and are more likely to exhibit negative neuroendocrine responses (Rodin & Timko, 1991). Thus, age-related differences in the sense of control pose a serious threat to health care systems serving rapidly aging populations.
Any relationship between age and the sense of control likely results from three sets of factors: (a) the increased volume and intensity of negative socially meaningful events; (b) the deterioration of functional and biomedical health status; and (c) the increased exposure to health professionals who prefer compliant, deferential, or institutionalized patients (Rodin, 1986a). Simply put, as older adults proceed through the life course, they have fewer control-enhancing experiences and encounter more control-restricting circumstances.
Schulz and Heckhausen (1999) identify three streams of relevant research: field experiments, large-scale epidemiological studies, and quasi-experimental studies. All three provide considerable evidence for age-related differences in the sense of control (Fung et al., 1999; Mirowsky, 1995; Rodin, 1986a, 1986b, 1987, 1990; Rodin & Timko, 1991; Rodin et al., 1985). That evidence, however, mostly comes from cross-sectional data, panel studies with short follow-up times, or surveys in which the sense of control is crudely measured (Andrisani, 1978; Lachman, 1985, 1986; McAvay, Seeman, & Rodin, 1996, Ross & Mirowsky, 2002).
Among the few existing longitudinal studies, the analysis by Pitcher, Spykerman, and Gazi-Tabatabaie (1987, p. 221) of the older male cohort of the National Longitudinal Study led them to conclude "the structure of personal control remains quite stable." Similarly, after reviewing the literature, Schulz, Heckhausen, and Locher (1991) concluded that there was little evidence that generalized beliefs about control change with advancing age, suggesting that observed age-related differences reflect cohort rather than aging effects. In contrast, Mirowsky (1995) reported a negative relationship between age and the sense of control in several cross-sectional state or national samples, consistently finding an accelerating decline that begins when older adults reach their fifth decade. A 3-year follow-up to one of those national samples shows that age is significantly related to declines over time in the sense of control, and that these declines are not notably diminished by the introduction of potential confounders (Ross & Mirowsky, 2002).
Thus, further research on the relationships among age, aging, and the sense of control is warranted. This is especially true given the limitations associated with Ross and Mirowsky's (2002) longitudinal study, which included few individuals over the age of 80, a 44% attrition rate, the narrow focus of the functional and sensory deficit measures, and important omitted potential confounders such as social support, stress, religiosity, and interactions with health personnel.
Accordingly, the purpose of this article is to report on a complementary longitudinal study. Mirowsky and Ross's (1991) sense of control measure was administered at baseline and at each of six bimonthly follow-up interviews with a large sample of chronically ill older adults. Reliable and valid measures of potential confounders of the relationship between age and the sense of control were obtained at baseline. These data allow us to examine the stability of the sense of control and the form of its relationships with age and aging, and they allow us to partition from those relationships any variance shared with theoretically relevant potential confounders.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
noted, Rodin (1986a) stressed the importance of studying the relationship between age and the sense of control among chronically ill older adults. We have access to data from a large, ongoing, longitudinal study of health-related quality of life (HRQoL) among older adults recruited from the general medicine and related outpatient clinics at the Indiana University School of Medicine, and the Saint Louis Veterans Administration Medical Center. With the use of electronic medical records, patients were identified as being potentially eligible based on age and medical criteria. Medical criteria for asthma (participants aged 18 years old or older), coronary artery disease or chronic heart failure (participants aged 50 years old or older), or chronic obstructive pulmonary disease (participants aged 50 years old or older) were specified by three expert panels of North American physicians. The 46 primary care physicians (PCPs) for these patients then reviewed the specific information for each of their patients and confirmed that the patients had the target diseases. Enrollment was limited to the 2,493 confirmed patients who kept scheduled visits during August 2000 to November 2001, of whom 1,662 (66.7%) were enrolled and interviewed at baseline.
Design
At the index visit, a brief face-to-face interview was conducted. It confirmed that a physician had told the patient that she or he had the target disease, verified demographic and telephone information, obtained a history of cigarette smoking and other conditions, and identified symptoms most troublesome to the patient. Within 3 days, patients received a baseline telephone interview conducted by Harris Interactive that included disease-specific and generic HRQoL measures, as well as several demographic, socioeconomic, and psychosocial indicators. Mean interview time was 33.7 min.
Harris Interactive conducted six telephone follow-up interviews over the next year. "Anniversary" follow-ups occurred at 2-month intervals. They could, however, occur earlier if the patient had a return visit to his or her PCP that was no more than 1 month before the next anniversary interview. If an early visit occurred, the clock was reset to the next anniversary follow-up. Therefore, all patients were scheduled to have six follow-up interviews, with 1–3 months between each visit (i.e., 2 months between two anniversary visits, at least 1 month between an anniversary and subsequent early visit, and no more than 3 months between an early and subsequent anniversary visit).
The patients' PCPs were also surveyed. At baseline, the PCP rated the seriousness of the patient's condition and the chances that the patient would be hospitalized or die within 2 years (two questions), and the PCP indicated whether the patient was known to be on medications, have had laboratory tests or procedures ordered, or been referred to a specialist for the target condition (three questions). The PCP response rate was 99.8%. If a patient follow-up interview occurred along with a visit to the PCP, then the PCP was asked six questions about how much the patient's condition had changed (three questions) and whether that change resulted in altering the medication regimen, ordering laboratory tests, or referring the patient to a specialist (three questions).
Sense of Control
Mirowsky and Ross's (1991) sense of control measure is conceptually similar to Rotter's (1966) but balances claiming versus denying responsibility. Patients agree or disagree with eight statements: 1, I am responsible for my own successes; 2, I can do just about anything I really set my mind to; 3, my misfortunes are the result of mistakes I have made; 4, I am responsible for my failures; 5, the really good things that happen to me are mostly luck; 6, there's no sense planning a lot—if something good is going to happen it will; 7, most of my problems are due to bad breaks; and 8, I have little control over the bad things that happen to me. Statements 1–4 reflect responsibility (coded -2, -1, 0, 1, and 2 for strongly disagree, disagree, don't know, agree, and strongly agree, respectively). Statements 5–8 reflect fatalism (and are reverse coded). Thus, this measure ranges from -16 to +16 (
; 
; 
).
The sense of control measure has strengths and weaknesses. On one hand, balancing the number of responsibility and fatalism statements eliminates the agreement bias associated with age and low education, improving the measure's validity. On the other hand, internal consistency-based reliability estimates are diminished, and exploratory factor analyses "mistakenly" identify responsibility and fatalism as separate factors (Mirowsky & Ross, 1991, 1996; Ross & Mirowsky, 2002). When a valence factor is used to represent the agreement bias, however, then a confirmatory factor analysis (Arbuckle & Wothke, 1999) of the baseline interviews (not shown) fits the data well (i.e., normed fit index =.926; incremental fit index =.932; root mean squared error of approximation =.082).
Age
Figure 1 graphs mean sense of control by actual age at baseline. The pattern is clear and tight between the ages of 45 and 85, which includes 93% of the sample. There is a modest negative relationship that appears to accelerate. Quadratic and cubed (as well as other) transformations of the number of years since reaching age 18 (
; 
) were evaluated, and, as in previous reports (Mirowsky, 1995; Mirowsky & Ross, 1991, 1996; Ross & Mirowsky, 2002), the cubed transformation fit the relationship best, although the improvement in fit was minimal. Thus, in the cross-sectional analyses the cubed transformation is used, and in the longitudinal analyses the quadratic transformation is used (Ross & Mirowsky, 2002).
|
). We consider four categories of potential confounders: demographic and socioeconomic status indicators, psychosocial factors, physical and mental well-being, and the patient–practitioner relationship. The demographic characteristics are gender and race. It is assumed that the culture in which these older adults have lived resulted in women having a lower sense of control because of the accumulative disadvantages experienced over the life course (Elder & Liker, 1982; O'Rand, 1996). Gender is measured with a binary variable coded 1 for men (61.1%). Minorities are also assumed to have a lower sense of control because of the accumulative disadvantages they have faced (Hughes & Demo, 1989; Mirowsky and Ross, 1990). Race is measured by a binary variable coded 1 for Whites (67.5%; there were no differences between Blacks, 28.4%, and others, 4.2%, on the sense of control).
Because gender and race differences are intertwined with socioeconomic status, education, employment status, and income are also included. Education leads to the development of abilities crucial to economic success and coping skills (Mirowsky & Ross, 1998; Pearlin, Menaghan, Lieberman, & Mullan, 1981). Education is measured in years of completed formal schooling (
; 
). Accomplishments associated with employment enhance the sense of control (Andrisani, 1978; Ross & Mirowsky, 1992). Employment status is measured with a set of three dummy variables reflecting working for pay (17.0%), being retired or unable to work (38.5%), and not having a substantial history of labor force participation (the reference group; 45.5%; note that any shared effect involving the inability to work should be eliminated by the introduction of the physical and mental well-being measures). Income provides access to resources that can be marshaled to maintain or enhance the sense of control in the face of adversity (Downey & Moen, 1987). We asked patients to subjectively assess whether their incomes were comfortable (23.4%), just enough to get by (50.5%), or were not even enough to get by (25.9%), using the latter as the reference group in a set of three dummy variables.
Psychosocial factors are the second category of potential confounders. The literature on stress, coping, and well-being is replete with evidence that stress is harmful, and that social support is protective (Pearlin et al., 1981). Stress is measured by a two-item scale from the 1990 NORC National Health Survey (McHorney & Lerner, 1991) that taps the frequency and intensity of stress. Summed scores from these two items were transformed such that 0 reflects maximal stress and 100 reflects minimal stress (
; 
; 
). Social support is measured by a five-item subset of the Medical Outcomes Study social support scale (Sherbourne & Stewart, 1991). Summed scores from these five items were transformed such that 0 reflects minimal social support and 100 reflects maximal social support (
; 
; 
). Religion is a crucial framing mechanism with clear etiologic linkages to health (Koening, McCullough, & Larson, 2001; Levin, 2001), and it likely serves to enhance and maintain the sense of control. The summary religiosity and spirituality items from the Fetzer instrument (Fetzer Institute, 1999) are used as a two-item scale transformed such that 0 reflects minimal religiosity and 100 reflects maximal religiosity (
; 
; 
).
The third category of potential confounders is physical and mental well-being. Despite being one of the most poorly measured domains, good health has consistently been associated with greater sense of control (Mirowsky, 1995). Because health declines with age, it is imperative to remove as much of this joint variance as possible. The Short Form 36 (SF-36, Version 2; Ware, Kosinski, & Dewey, 2000) is the most widely used HRQoL measure. In this analysis, we use the physical composite score (PCS) and the mental composite score (MCS) because these maximize granularity. Both the PCS (
; 
) and MCS (
; 
) have been transformed such that 0 reflects the worst well-being and 100 reflects the best well-being.
Patient–practitioner relationships are the last category of potential confounders. Although exposure to health professionals is one of the factors thought to account for the relationship between age and the sense of control (Rodin, 1986a), it has seldom been investigated. We use the 10-item patient satisfaction scale developed by the American Board of Internal Medicine (ABIM; Webster, 1989). The ABIM addresses the affective nature of the patient–practitioner relationship, with 0 reflecting the least satisfaction and 100 reflecting the most (
; 
; 
). The horizon hypothesis (Mirowsky, 1997) assumes that greater subjective life expectancy over the life course results in higher levels of the sense of control. Because life expectancy is negatively associated with age, it must also be partitioned. In the absence of the patient's subjective estimate of her or his life expectancy, we use the PCP's subjective estimate of the percentage chance that the patient would die within 2 years (
; 
).
Analytic Approach
There are three phases to the analysis. The first is a descriptive examination of level and normative stability (Mortimer, Finch, & Kumka, 1982). Level stability considers the similarity of group means over time, and normative stability focuses on the correlation of scores over time. We also use repeated measures analysis of covariance. These descriptive analyses are done for all patients enrolled in the study, as well as for the subset of patients who have completed all follow-up interviews. The subgroup analyses examine the potential for attrition bias, which is likely to be rather small (Allison, 2002). Of the 1,063 patients whose 1-year enrollment ended by June 1, 2002, 691 (64.3%) had completed all six follow-up interviews, and 909 (85.5%) had completed either the fourth, fifth, or sixth follow-up interview.
The second phase of the analysis involves a hierarchical multiple linear regression of the baseline sense of control measure to evaluate the static association between age and the sense of control. Age is entered into the first model, with the next four models sequentially introducing one category of potential confounders at a time. Ordinary least squares (OLS) regression is used, and the OLS assumptions are evaluated by use of standard methods (Allison, 1999).
In the third phase of the analysis, hierarchical multiple residualized change score regression is used (Kessler & Greenberg, 1981). To minimize attrition bias, the dependent variable is the sense of control measured at each patient's last (as of June 1, 2002) follow-up visit. The baseline sense of control measure is entered into the first model by itself. The number of the patient's last follow-up interview is entered into the second model to adjust for exposure differences. Age is added to the third model, with the next four models sequentially introducing one category of the potential confounders at a time. OLS regression is used, and the OLS assumptions are evaluated by use of standard methods (Allison, 1999).
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; 
). Variability in the sense of control shows a similar pattern of increase over time, with the exception of the fifth follow-up.
|
; 
).
|
) significant. The smallest is.527, with the six wave-to-wave correlations appreciably higher (i.e.,.578,.641,.686,.700,.691, and.747, respectively). Wave-to-wave paired sample t tests (not shown) indicate that the increases in the sense of control over time were not statistically significant for any 1- to 3-month interval (
). Similar results were obtained from a repeated measures analysis of covariance of these data. Those results (also not shown) revealed no meaningful within-subject effects among the potential confounders. Thus, although there is an aggregate tendency for patients to claim greater responsibility over time, the amount of change that occurs over any 1- to 3-month period is neither significant nor predictable.
|
).
|
; 
), and it accounts for only 0.4% of the variance. The size of the relationship increases appreciably (
; ) with the introduction of the demographic and socioeconomic status indicators, increases marginally (
; ) with the introduction of the psychosocial factors, and is then impervious to the introduction of both the physical and mental well-being, and the patient–practitioner measures. No gender differences are observed. Whites have statistically significant but modestly higher scores than non-Whites. There is a significant positive association with education. Current and prior labor force participation is equally beneficial, compared with never having worked for pay. A dose-response, positive relationship with income is observed, but it is eliminated by the introduction of physical and mental well-being. The psychosocial measures have significant positive associations when they are first introduced, but these are also eliminated by the introduction of the physical and mental well-being measures. Both physical and mental well-being have positive, statistically significant associations, with the former being modest (
; ) and the latter being substantial (
; ). The patient–practitioner measures are not associated with the sense of control.
Multivariable Longitudinal Analysis
The preceding analyses have shown two things. First, there is an aggregate but insignificant and unpredictable wave-to-wave trend toward an increase in the sense of control across follow-ups. Second, there is a statistically significant, modestly negative association between age and the sense of control at baseline that is not accounted for by the introduction of the potential confounders. This begs the question of whether significant change in the sense of control occurs over the entire 1-year follow-up period, and, if so, whether age or the potential confounders are associated with it.
As a way to address this, a multivariable residualized change score regression analysis (Kessler & Greenberg, 1981) of the 1,587 patients (95.5%) who completed one or more follow-up interviews was conducted. The dependent variable is the sense of control measured at the patient's last follow-up. This raises a potential exposure or censoring issue because the amount of time is not constant across all patients, and because there is a clear, albeit nonmonotonic, dose-response relationship between elapsed time and changes in the sense of control. Therefore, we include an indicator of elapsed time. This results in a seven-stage hierarchical model that sequentially introduces the sense of control at baseline, the elapsed time indicator, age, and the four categories of potential confounders. In these longitudinal analyses, age is coded as the squared value of the number of years since reaching age 18 (Ross and Mirowsky, 2002).
Table 5 contains the results of the multivariable residualized change score regression analysis. Again, all assumptions (including multicollinearity) of the residualized change score regression model were assessed by using standard methods, and no violations were detected (Allison, 1999). The effects of age and the potential confounders on changes in the sense of control are reflected in the betas associated with those measures, whereas stability in the sense of control over time is reflected in the value of the unstandardized regression coefficient, b, associated with the sense of control at baseline (Kessler & Greenberg, 1981). If the b coefficient is unity, then no structured change has occurred (although random changes at the individual level that sum to zero in the aggregate may be observed); if b is greater than unity, then there is structured change (on average, everyone's score went up an amount proportionally equivalent to the b coefficient minus unity); and, if that change is less than unity, then there is regression to the mean.
|
; 
; ) for the sense of control at baseline reflect the considerable stability of that measure, as well as regression to the mean. This crude effect is only marginally diminished by the introduction of age and the potential confounders (
; 
; ). Exposure time has a significant, but modest positive effect on the amount of change. Age has a statistically significant but modest negative association (
; ) when first introduced, but it is buoyed by the introduction of the demographic and socioeconomic factors (
; ), as well as by the psychosocial factors (
; ). In the final model, the moderate effect of age is larger than that for any factor aside from the sense of control at baseline.
Gender is not associated with changes in the sense of control. Modest yet enduring and statistically significant increases occur for Whites compared with non-Whites and for better as opposed to less educated patients. Having a more comfortable income is initially associated with modest increases in the sense of control, but this advantage is eliminated after physical and mental well-being are adjusted for. All three of the psychosocial factors are initially moderately positively associated with changes in the sense of control, but the association with stress is eliminated when physical and mental well-being are added to the model. Both physical and mental well-being have significant, positive associations, with the former being modest and the latter being more moderate. Indeed, aside from the effects of age and of the baseline value of the sense of control, the association with mental well-being is the most substantial (
; ). The patient–practitioner relationship measures are not associated with changes in the sense of control.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Lachman, 1985, 1986; McAvay et al., 1996; Ross & Mirowsky, 2002) of negative study results—age-related changes in the sense of control may not have been found because the decline is gradual.
A second limitation is that only chronically ill patients with asthma, coronary artery disease or chronic heart failure, or chronic obstructive pulmonary disease were studied. For some, this will raise concerns about external validity. These three diseases, however, are rather common among older adults (Department of Health and Human Services, 1991). Moreover, focusing on chronically ill older adults is entirely consistent with extant theory (Rodin, 1986a). Indeed, the focus on chronically ill older adults may have facilitated the detection of changes in the sense of control over this brief follow-up period, because such individuals are most likely to experience control-restricting experiences, especially those associated with the natural course of their health conditions and associated treatment regimens.
Nonetheless, to address the potential that age may simply be taking credit for the natural progressive course of these diseases, we conducted additional sensitivity analyses. This involved estimation of an eighth model for Table 5 that introduced measures of the changes between baseline and the last follow-up interview in the SF-36 10-item physical function subscale score and the SF-36 5-item mental health subscale score. We chose to introduce changes in these subscale scores rather than changes in the PCS and MCS themselves in order to minimize multicollinearity problems. The results (not shown) demonstrated that although the R2 increased by 2.2% and both of the changes in subscale scores were statistically significant, the parameter estimates obtained for age and the potential confounders were equivalent to those shown in the seventh model in Table 5. Thus, we are further convinced that our results are not an artifact of the focus on a chronically ill sample of older adults.
Concerns about external validity are also reflected in the third limitation. Patients came from only two medical centers, and at the Veterans Administration facility, the vast majority of patients were men. To address this, we conducted sensitivity analyses using only the Indiana University medical center patients. Those results (not shown) were consistent with our reported findings, including the absence of any cross-sectional or longitudinal association between gender and the sense of control. Moreover, Indiana University medical center patients came from six geographically dispersed outpatient clinic sites serving individuals from all walks of life.
The fourth limitation is that the horizon hypothesis was examined indirectly with a crude measure. That is, subjective assessments of patient life expectancies were obtained from the PCPs, who rated the percentage chances of dying as 10% or less for 73.2%. As a way to address this, sensitivity analyses were conducted by using a binary marker contrasting those in the upper quartile of risk versus all others. Those results (not shown) were essentially the same. Nonetheless, our inability to replicate Mirowsky's (1997) previous report that the sense of control is negatively associated with life expectancy may be due to reliance on PCP rather than patient assessments.
Similarly, the fifth limitation involves the crude measurement of the patient–practitioner relationship. Although the ABIM measure of patient satisfaction is used widely in research and in the evaluation of medical training, it is a somewhat oblique measure for the increased exposure to health professionals who prefer compliant, deferential, or institutionalized patients. Thus, our inability to find an association between the patient–practitioner relationship and the sense of control, both cross-sectionally and longitudinally, may be due to measurement error.
A sixth limitation is that the analyses rely on a static assessment of the relationship among change in the sense of control, age, and the potential confounders. Our models do not consider changes in the predictor variables from baseline. Finally, although the data brought to bear in these analyses are extensive, not all of the patients had an opportunity to complete their final follow-up interview. Although the longitudinal and sensitivity analyses address this potential for bias, a definitive resolution is not possible at this time.
These limitations notwithstanding, this study makes three important contributions. First, there is compelling evidence that a statistically and substantively significant negative relationship exists between age and the sense of control, both cross-sectionally and longitudinally. As Mirowsky (1995; Ross & Mirowsky, 2002) has suggested, the functional form is not linear but is best captured cross-sectionally by the cubed value of the number of years since reaching age 18, and longitudinally by the squared value. These functional forms reflect the accelerating decline associated with age and aging, respectively. Second, although the negative association between age and the sense of control is moderate, it has the second most robust relationship in the cross-sectional analyses, and the most robust relationship in the longitudinal analyses (aside from the sense of control at baseline). Third, in both the cross-sectional and longitudinal analyses, mental well-being is the only other factor principally associated with the sense of control. Better mental well-being is associated with higher baseline values and greater improvements in it over time.
Where does that leave us? These findings have two major implications for future research on and conceptualization about the relationship between age and the sense of control. One involves a fundamental reorientation of basic research strategies and designs. As indicated earlier, most extant studies have either been cross-sectional or involved only one or two follow-up interviews. Moreover, the intervals between those follow-ups have been relatively short. Our study has shown that changes in the sense of control are gradual, and not entirely monotonic, let alone linear. Therefore, the field requires longitudinal studies with repeated assessments of the sense of control over a prolonged observation period. Only then can rigorous methods (i.e., latent growth curve models) be brought to bear to elucidate the complicated and embedded causal nexus of trajectories in the sense of control. It should be trajectories that researchers focus on in the future, rather than the simple transitions between one state and another.
The other major implication involves the substantial association between mental well-being and the sense of control. Previous studies have focused primarily on physical well-being. Although that effect was also shown in these data, it is clear that the cognitive interpretation and integration of one's health evaluations is a more salient correlate, even though it is influenced by physical well-being. Indeed, the role of mental well-being rivals that of age in magnitude. Therefore, a conceptual reconsideration of the broader psychosocial domain in which the sense of control resides is also needed. It will likely be fruitful to draw upon extant models of other cognitive constructs, such as intellectual abilities, cognitive speed, and memory.
Having stated the limitations of the study and made the case for its major findings and their implications, we find it appropriate to reconsider the contributions of this study in context. This work has taken important steps forward in evaluating the association among age, aging, and the sense of control. Those steps include the use of longitudinal data, the incorporation of up to six waves of follow-up interviews, and a careful examination of a larger but nonetheless incomplete set of potential confounders. Certainly, this is a worthy advance to the extant literature. At the same time, however, this study falls noticeably short of the kind of research that is called for. That having been said, let us not lose sight of the major pragmatic reason for studying the relationship between age and the sense of control in the first place. As Rodin and Timko (1991) have written, individuals with a lower sense of self-control take less responsibility for their health, are less likely to engage in health protective behaviors, have inhibited immunologic function, and are more likely to exhibit negative neuroendocrine responses. If age-related differences in the sense of control result because older adults have fewer control-enhancing experiences and encounter more control-restricting circumstances, then society should endeavor to buffer the adverse effects of those deleterious experiences and circumstances in order to enhance the sense of control in older adults and minimize health care costs.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication September 3, 2002. Accepted for publication January 24, 2003.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Index. Social Psychology Quarterly, 54,127-145.This article has been cited by other articles:
 |
|
 |
|
  H.-S. Jeon and R. E. Dunkle Stress and Depression Among the Oldest-Old: A Longitudinal Analysis Research on Aging, November 1, 2009; 31(6): 661 - 687. [Abstract] [PDF]
|
|
|
|
 |
|
 T. M. Gadalla Sense of Mastery, Social Support, and Health in Elderly Canadians J Aging Health, June 1, 2009; 21(4): 581 - 595. [Abstract] [PDF]
|
|
|
|
 |
|
 J. Cairney and N. Krause Negative Life Events and Age-Related Decline in Mastery: Are Older Adults More Vulnerable to the Control-Eroding Effect of Stress? J. Gerontol. B. Psychol. Sci. Soc. Sci., May 1, 2008; 63(3): S162 - S170. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. D. Windsor, K. J. Anstey, P. Butterworth, M. A. Luszcz, and G. R. Andrews The Role of Perceived Control in Explaining Depressive Symptoms Associated With Driving Cessation in a Longitudinal Study Gerontologist, April 1, 2007; 47(2): 215 - 223. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Krause Age and Decline in Role-Specific Feelings of Control J. Gerontol. B. Psychol. Sci. Soc. Sci., January 1, 2007; 62(1): S28 - S35. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
